The goal of an online survey is to gather quantitative primary research that is used to improve decisions related to identifying and solving problems.
Guiding principles of designing online surveys: Strive to make the survey reliable, accurate and consistent. That means the results should be replicable, and the sample population should be representative.
Many books have been written on this marketing tool, so I won’t pretend to cover everything in this post, but here are some basic pointers I’ve picked-up from experience:
1. Know your objective and stick to it. Online surveys are not the place for exploratory or feasibility research. They are best suited for service or product review, SERVQUAL topics, and specific feedback. Limit your questions to what is relevant to the marketing situation at hand, and consider your online demographic.
2. Keep it short and sweet. In my opinion, some of the best online surveys have been around for a rather long time. Weaker questions have already been weeded out, and the questions left are highly reliable for their constructs (sets of questions that measure a specific category). If you are designing a new survey, it is a good idea to begin by looking around the web for surveys that measure similar constructs to what you want to measure. Look at the length of the surveys and the clarity of wording for the questions for inspiration. A rule of thumb is that your online survey should have no more than 5 questions per construct, and no more than 20 questions.
The sweet part of rule 2 means make it as painless as possible for your participants. Avoid using these words: and, but, because, not. More complicated questions or personal questions should be kept at the end of the questionnaire. Also be careful not to use leading questions or bias answers with weighted choices.
3. Before it goes live, check the validity of your online survey. Here are a few basic techniques:
Face Validity- Recruit people (preferably members of the survey population) to read the survey questions and “back translate” the meaning of the question to you in their own words.
Convergent Validity- After administering the online survey to a test population, use two different methods to test validity such as Cronbach’s Alpha for item-total correlation, coupled with an alternative-forms test.
Discriminant Validity- Each set of questions is designed to measure a specific thing. The set should measure that one thing well and other things poorly. Using SPSS or other data analysis program, check the item-total correlation to identify weak questions.
4. In my opinion, the biggest hurdle for online surveys is the non-response bias. Because the sample size affects the accuracy of the results, it is important to remember when drawing conclusions from a sample that there is always a degree of error involved in sample selection. That is why it is important to motivate the survey’s audience to participate in the survey.
Most non-marketers usually assume this means money. While incentives can help motivate participants, they do not need to be of a monetary nature. For instance, access to information or membership status can be ample motivation for most surveys. That said, I have found that the best incentive is actually personal interest in the survey subject matter. Consider the large volume of popular surveys on Facebook.
If you are using an email list as your sample population, it helps to send reminders at appropriate intervals, including a final reminder. Many online survey platforms such as SurveyMonkey allow you to send reminders to your sample population who have yet to take the survey. If you are working from a website then it can be helpful to indicate a time limit for survey participation.
5. Don’t forget the personal touch. It is important to let participants know that their time and opinions matter. A personal invitation to participate can go a long way, as well as a thank you when they complete the survey.
Well, I think that covers the basics of the front end of the survey, your comments are welcome! The most exciting part of surveys is in the analysis, which will have to wait for another post.